Camunda simplifies process automation with new AI-powered natural language features

Process orchestration company Camunda has announced new artificial intelligence (AI) capabilities it plans to add to its business orchestration platform later this year.

Announced at its CamundaCon 2024 event in Berlin this week, the company’s pitch for AI is that the technology will hugely simplify the steep learning curve that comes with the setup phase of many business orchestration projects.

Most of the innovations are part of a new generative AI help system called Camunda Copilot, which the company said will simplify the many complex tasks required for process orchestration.

One example it offered is that employees often struggle with basic elements such as translating their understanding of what is required for process orchestration into the industry standard graphical language, Business Process Modeling Notation (BPMN).

However, using a feature called Copilot Suggestions, an employee will be able to describe the process they are trying to model in natural language and have this translated by the AI into BPMN.

“The problem that you’re seeing is a skill gap. You have people who know what a process is, but they can’t turn it into a BPMN file because they lack the knowledge,” Camunda CTO Daniel Meyer told CIO.  

Similar AI assistance has been available since last year in the platform’s AI-assisted Form Builder, the component of the Camunda platform used to create many different types of form used for everyday business workflow. Using this, the developer can describe the form they want to create for the task and the AI generates it for them.

“This allows us to lower the skill gap for who can create such processes and what they need to know,” said Meyer. “In Form Builder, this allows you to create forms using natural language.”

The company is also making available AI connectors that organizations could use to integrate the Camunda platform with generative AI machine learning platforms from OpenAI, Azure OpenAI, and Hugging Face without the need for a lot of experience or development resources, he said.

“There are decisions where we need a human, but we can augment this with AI. And then there are tasks where we’ve had humans work on them in the past but where they could be working on other stuff.”

The AI effect

Process orchestration is sometimes seen as the dry preserve of business architects and yet getting it right has become fundamental to every digital business.

An example used by Camunda is ordering a product to pick up in person from a large online retailer. Behind this relatively simple transaction lie multiple processes that require the coordination of different software systems, human decision making, layers of communication, and the prioritization of different events.

If any part of this chain breaks down, the product won’t be ready for the customer, or will be ready but they won’t have been told to collect it at the right time.

Adding to this is the challenge of legacy technology and processes that can be difficult to integrate into modern workflows.

It is in this context that AI is turning up on specification sheets as the solution, sometimes reframing innovations that have been around for some time.

This has led to big claims that run the risk of hype and disappointment down the line. According to Forrester analyst Bernhard Schaffrik, Camunda’s announcement is more evolutionary, promoting AI as a technology customers could use today as an orchestration helpmate.

“There are vendors who market AI as a revolution that will replace any of the technologies they have built so far to automate tasks and processes, which is something I don’t fully buy into,” he said.  

The alternative is a more pragmatic approach that emphasizes how AI could lower the entry barriers to using process orchestration. 

“Camunda positions AI as a helper, augmenting existing capabilities and lowering the entry barriers for people to benefit from the technology,” he said.

Schaffrik sees the challenge AI poses as one of implementation. Organizations lack the skills to implement AI right now. Compliance is another issue.

“Unless you have a solid understanding of what you would like to achieve with process automation, [using] AI all over the place doesn’t necessarily help,” he said.

The new AI features are currently being rolled out experimentally to some customers, Camunda said. They will be added to Camunda 8.5 more generally over the next six months.